Module

4
Analysis of Statically
Indeterminate
Structures by the Direct
Stiffness Method
Version 2 CE IIT, Kharagpur
 Lesson
23
The Direct Stiffness
Method: An
Introduction
Version 2 CE IIT, Kharagpur
Instructional Objectives:
After reading this chapter the student will be able to
1. Differentiate between the direct stiffness method and the displacement
method.
2. Formulate flexibility matrix of member.
3. Define stiffness matrix.
4. Construct stiffness matrix of a member.
5. Analyse simple structures by the direct stiffness matrix.

23.1 Introduction
All known methods of structural analysis are classified into two distinct groups:-

(i) force method of analysis and

(ii) displacement method of analysis.

In module 2, the force method of analysis or the method of consistent

deformation is discussed. An introduction to the displacement method of analysis
is given in module 3, where in slope-deflection method and moment- distribution
method are discussed. In this module the direct stiffness method is discussed. In
the displacement method of analysis the equilibrium equations are written by
expressing the unknown joint displacements in terms of loads by using load-
displacement relations. The unknown joint displacements (the degrees of
freedom of the structure) are calculated by solving equilibrium equations. The
slope-deflection and moment-distribution methods were extensively used before
the high speed computing era. After the revolution in computer industry, only
direct stiffness method is used.

The displacement method follows essentially the same steps for both statically
determinate and indeterminate structures. In displacement /stiffness method of
analysis, once the structural model is defined, the unknowns (joint rotations and
translations) are automatically chosen unlike the force method of analysis.
Hence, displacement method of analysis is preferred to computer
implementation. The method follows a rather a set procedure. The direct stiffness
method is closely related to slope-deflection equations.

The general method of analyzing indeterminate structures by displacement

method may be traced to Navier (1785-1836). For example consider a four
member truss as shown in Fig.23.1.The given truss is statically indeterminate to
second degree as there are four bar forces but we have only two equations of
equilibrium. Denote each member by a number, for example (1), (2), (3) and (4).
Let α i be the angle, the i-th member makes with the horizontal. Under the

Version 2 CE IIT, Kharagpur

action of external loads Px and Py , the joint E displaces to E’. Let u and v be its
vertical and horizontal displacements. Navier solved this problem as follows.

In the displacement method of analysis u and v are the only two unknowns for
this structure. The elongation of individual truss members can be expressed in
terms of these two unknown joint displacements. Next, calculate bar forces in the
members by using force–displacement relation. Now at E, two equilibrium
equations can be written viz., ∑F x = 0 and ∑F y = 0 by summing all forces
in x and y directions. The unknown displacements may be calculated by solving
the equilibrium equations. In displacement method of analysis, there will be
exactly as many equilibrium equations as there are unknowns.

Let an elastic body is acted by a force F and the corresponding displacement be

u in the direction of force. In module 1, we have discussed force- displacement
relationship. The force (F) is related to the displacement (u) for the linear elastic
material by the relation
F = ku (23.1)

where the constant of proportionality k is defined as the stiffness of the structure

and it has units of force per unit elongation. The above equation may also be
written as

u = aF (23.2)

Version 2 CE IIT, Kharagpur

The constant a is known as flexibility of the structure and it has a unit of
displacement per unit force. In general the structures are subjected to n forces at
n different locations on the structure. In such a case, to relate displacement at i
to load at j , it is required to use flexibility coefficients with subscripts. Thus the
flexibility coefficient a ij is the deflection at i due to unit value of force applied at
j . Similarly the stiffness coefficient k ij is defined as the force generated at i

Version 2 CE IIT, Kharagpur

due to unit displacement at j with all other displacements kept at zero. To
illustrate this definition, consider a cantilever beam which is loaded as shown in
Fig.23.2. The two degrees of freedom for this problem are vertical displacement
at B and rotation at B. Let them be denoted by u1 and u 2 (= θ 1 ). Denote the
vertical force P by P1 and the tip moment M by P2 . Now apply a unit vertical
force along P1 and calculate deflection u1 and u 2 .The vertical deflection is
denoted by flexibility coefficient a11 and rotation is denoted by flexibility
coefficient a 21 . Similarly, by applying a unit force along P1 , one could calculate
flexibility coefficient a12 and a 22 . Thus a12 is the deflection at 1 corresponding to
P1 due to unit force applied at 2 in the direction of P2 . By using the principle of
superposition, the displacements u1 and u 2 are expressed as the sum of
displacements due to loads P1 and P2 acting separately on the beam. Thus,

u1 = a11 P1 + a12 P2
u 2 = a 21 P1 + a 22 P2 (23.3a)

The above equation may be written in matrix notation as

{u} = [a]{P}
⎧u ⎫ ⎡ a11 a12 ⎤ ⎧P ⎫
where, {u} = ⎨ 1 ⎬ ; {a} = ⎢ ⎥ ; and { P} = ⎨ 1 ⎬
⎩u2 ⎭ ⎣ a21 a22 ⎦ ⎩ P2 ⎭

Version 2 CE IIT, Kharagpur

The forces can also be related to displacements using stiffness coefficients.
Apply a unit displacement along u1 (see Fig.23.2d) keeping displacement u 2 as
zero. Calculate the required forces for this case as k11 and k 21 .Here, k 21
represents force developed along P2 when a unit displacement along u1 is
introduced keeping u 2 =0. Apply a unit rotation along u 2 (vide Fig.23.2c) ,keeping
u1 = 0 . Calculate the required forces for this configuration k12 and k 22 . Invoking
the principle of superposition, the forces P1 and P2 are expressed as the sum of
forces developed due to displacements u1 and u 2 acting separately on the beam.
Thus,

P1 = k11u1 + k12 u 2

P2 = k 21u1 + k 22 u 2 (23.4)

{P} = [k ]{u}
Version 2 CE IIT, Kharagpur
 ⎧P ⎫ ⎡k k12 ⎤ ⎧u ⎫
where, { P} = ⎨ 1 ⎬ ; {k} = ⎢ 11 ⎥ ;and {u} = ⎨ 1 ⎬ .
⎩ P2 ⎭ ⎣ k21 k22 ⎦ ⎩u2 ⎭

[k ] is defined as the stiffness matrix of the beam.

In this lesson, using stiffness method a few problems will be solved. However this
approach is very rudimentary and is suited for hand computation. A more formal
approach of the stiffness method will be presented in the next lesson.

23.2 A simple example with one degree of freedom

Consider a fixed–simply supported beam of constant flexural rigidity EI and span
L which is carrying a uniformly distributed load of w kN/m as shown in Fig.23.3a.

If the axial deformation is neglected, then this beam is kinematically

indeterminate to first degree. The only unknown joint displacement is θ B .Thus
the degrees of freedom for this structure is one (for a brief discussion on degrees
of freedom, please see introduction to module 3).The analysis of above structure
by stiffness method is accomplished in following steps:

1. Recall that in the flexibility /force method the redundants are released (i.e.
made zero) to obtain a statically determinate structure. A similar operation
in the stiffness method is to make all the unknown displacements equal to
zero by altering the boundary conditions. Such an altered structure is
known as kinematically determinate structure as all joint displacements
are known in this case. In the present case the restrained structure is
obtained by preventing the rotation at B as shown in Fig.23.3b. Apply all
the external loads on the kinematically determinate structure. Due to
restraint at B, a moment M B is developed at B. In the stiffness method we
adopt the following sign convention. Counterclockwise moments and
counterclockwise rotations are taken as positive, upward forces and
displacements are taken as positive. Thus,

wl 2
MB = − (-ve as M B is clockwise) (23.5)
12

The fixed end moment may be obtained from the table given at the end of lesson
14.
2. In actual structure there is no moment at B. Hence apply an equal and
opposite moment M B at B as shown in Fig.23.3c. Under the action of (-
M B ) the joint rotates in the clockwise direction by an unknown amount. It
is observed that superposition of above two cases (Fig.23.3b and
Fig.23.3c) gives the forces in the actual structure. Thus the rotation of joint
Version 2 CE IIT, Kharagpur
 B must be θ B which is unknown .The relation between − M B and θ B is
established as follows. Apply a unit rotation at B and calculate the
moment. ( k BB ) caused by it. That is given by the relation

4 EI
k BB = (23.6)
L

where k BB is the stiffness coefficient and is defined as the force at joint B due to
unit displacement at joint B. Now, moment caused by θ B rotation is

M B = k BBθ B (23.7)

3. Now, write the equilibrium equation for joint B. The total moment at B is
M B + k BBθ B , but in the actual structure the moment at B is zero as
support B is hinged. Hence,

M B + k BBθ B = 0 (23.8)

MB
θB = −
k BB

wl 3
θB = (23.9)
48 EI

4 EI
The relation M B = θ B has already been derived in slope –deflection method
L
in lesson 14. Please note that exactly the same steps are followed in slope-
deflection method.

Version 2 CE IIT, Kharagpur

Version 2 CE IIT, Kharagpur
23.3 Two degrees of freedom structure
Consider a plane truss as shown in Fig.23.4a.There is four members in the truss
and they meet at the common point at E. The truss is subjected to external loads
P and P acting at E. In the analysis, neglect the self weight of members. There
1 2
are two unknown displacements at joint E and are denoted by u 1 and u 2 .Thus
the structure is kinematically indeterminate to second degree. The applied forces
and unknown joint displacements are shown in the positive directions. The
members are numbered from (1), (2), (3) and (4) as shown in the figure. The
length and axial rigidity of i-th member is l and EAi respectively. Now it is sought
i
to evaluate u 1 and u 2 by stiffness method. This is done in following steps:

1. In the first step, make all the unknown displacements equal to zero by
altering the boundary conditions as shown in Fig.23.4b. On this restrained
/kinematically determinate structure, apply all the external loads except
the joint loads and calculate the reactions corresponding to unknown joint
displacements u 1 and u 2 . Since, in the present case, there are no
external loads other than the joint loads, the reactions ( R L )1 and ( R L ) 2
will be equal to zero. Thus,

⎧( R L )1 ⎫ ⎧0⎫
⎨ ⎬=⎨ ⎬ (23.10)
⎩( R L ) 2 ⎭ ⎩0⎭

2. In the next step, calculate stiffness coefficients k11 , k 21 , k12 and k 22 .This is
done as follows. First give a unit displacement along u1 holding
displacement along u 2 to zero and calculate reactions at E corresponding
to unknown displacements u1 and u 2 in the kinematically determinate
structure. They are denoted by k11 , k 21 . The joint stiffness k11 , k 21 of the
whole truss is composed of individual member stiffness of the truss. This
is shown in Fig.23.4c. Now consider the member AE . Under the action of
unit displacement along u1 , the joint E displaces to E ′ . Obviously the new
length is not equal to length AE . Let us denote the new length of the
members by l1 + Δl1 , where Δl , is the change in length of the
member AE ′ . The member AE ′ also makes an angle θ 1 with the
horizontal. This is justified as Δl1 is small. From the geometry, the change
in length of the members AE ′ is

Δl1 = cos θ 1 (23.11a)

Version 2 CE IIT, Kharagpur

The elongation Δl1 is related to the force in the member AE ′ , FAE' by

F AE ' l1
Δl1 = (23.11b)
A1 E

Thus from (23.11a) and (23.11b), the force in the members AE ′ is

EA1
′ =
FAE cos θ1 (23.11c)
l1

This force acts along the member axis. This force may be resolved along u1 and
EA1
′ is
u 2 directions. Thus, horizontal component of force FAE cos 2 θ 1 (23.11d)
l1
EA1
′ is
and vertical component of force FAE cos θ 1 sin θ 1 (23.11e)
l1

Version 2 CE IIT, Kharagpur

Version 2 CE IIT, Kharagpur
Version 2 CE IIT, Kharagpur
Expressions of similar form as above may be obtained for all members. The sum
of all horizontal components of individual forces gives us the stiffness coefficient
k11 and sum of all vertical component of forces give us the required stiffness
coefficient k 21 .

EA1 EA2 EA3 EA4

k11 = cos 2 θ 1 + cos 2 θ 2 + cos 2 θ 3 + cos 2 θ 4
l1 l2 l3 l4

4
EAi
k11 = ∑ cos 2 θ i (23.12)
i =1 li

∑
EAi
k 21 = cos θ i sin θ i (23.13)
li

In the next step, give a unit displacement along u 2 holding displacement along
u1 equal to zero and calculate reactions at E corresponding to unknown
displacements u1 and u 2 in the kinematically determinate structure. The
corresponding reactions are denoted by k12 and k 22 as shown in Fig.23.4d. The
joint E gets displaced to E ′ when a unit vertical displacement is given to the joint
as shown in the figure. Thus, the new length of the member AE ′ is l1 + Δl1 .
From the geometry, the elongation Δl1 is given by

Δl1 = sin θ 1 (23.14a)

EA1
Thus axial force in the member along its centroidal axis is sin θ1 (23.14b)
l1

Resolve the axial force in the member along u1 and u 2 directions. Thus,
EA1
horizontal component of force in the member AE ′ is sin θ 1 cos θ 1 (23.14c)
l1
EA1
and vertical component of force in the member AE ′ is sin 2 θ 1 (23.14d)
l1

In order to evaluate k 22 , we need to sum vertical components of forces in all the

members meeting at joint E .Thus,

Version 2 CE IIT, Kharagpur

 4

∑
EAi
k 22 = sin 2 θ i (23.15)
i =1
li

∑
EAi
Similarly, k12 = sin θ i cos θ i (23.16)
i =1
li

3. Joint forces in the original structure corresponding to unknown

displacements u1 and u 2 are

⎧ F1 ⎫ ⎧ P1 ⎫
⎨ ⎬=⎨ ⎬ (23.17)
⎩ F2 ⎭ ⎩ P2 ⎭
Now the equilibrium equations at joint E states that the forces in the original
structure are equal to the superposition of (i) reactions in the kinematically
restrained structure corresponding to unknown joint displacements and (ii)
reactions in the restrained structure due to unknown displacements themselves.
This may be expressed as,

F1 = (R L )1 + k11u1 + k12 u 2
F2 = (R L )2 + k 21u1 + k 22 u 2 (23.18)

This may be written compactly as

{F } = {Ri } + [k ]{u} (23.19)

where,

⎧ F1 ⎫
{F } = ⎨ ⎬;
⎩ F2 ⎭

⎧(R ) ⎫
{R L } = ⎨ L 1 ⎬
⎩(R L )2 ⎭

[k ] = ⎡⎢
k11 k12 ⎤
⎣k 21 k 22 ⎥⎦

⎧u1 ⎫
{u} = ⎨ ⎬ (23.20)
⎩u 2 ⎭

Version 2 CE IIT, Kharagpur

 L
For example take P1 = P2 = P , Li = , A1 = A2 = A3 = A4 = A and
sin θ i
θ 1 = 35° , θ 2 = 70° , θ 3 = 105° and θ 4 = 140°

Then.

⎧P ⎫
{F } = ⎨ ⎬ (23.21)
⎩P ⎭

{R L } = ⎧⎨
0⎫
⎬
⎩0⎭

∑
EA EA
k11 = cos 2 θ i sin θ i = 0.9367
L L

∑
EA EA
k12 = sin 2 θ i cos θ i = 0.0135
L L

∑
EA EA
k 21 = sin 2 θ i cos θ i = 0.0135
L L

∑
EA EA
k 22 = sin 3 θ i = 2.1853 (23.22)
L L

⎧ P ⎫ EA ⎡0.9367 0.0135⎤ ⎧ u1 ⎫
⎨ ⎬= ⎢ ⎥⎨ ⎬
⎩ P ⎭ L ⎣ 0.0135 2.1853⎦ ⎩u2 ⎭

Solving which, yields

L
u1 = 1.0611
EA
L
u 2 = 0.451
EA

Version 2 CE IIT, Kharagpur

Example 23.1
Analyze the plane frame shown in Fig.23.5a by the direct stiffness method.
Assume that the flexural rigidity for all members is the same .Neglect axial
displacements.

Solution
In the first step identify the degrees of freedom of the frame .The given frame has
three degrees of freedom (see Fig.23.5b):

(i) Two rotations as indicated by u1 and u 2 and

(ii) One horizontal displacement of joint B and C as indicated by u 3 .

In the next step make all the displacements equal to zero by fixing joints B and C
as shown in Fig.23.5c. On this kinematically determinate structure apply all the
external loads and calculate reactions corresponding to unknown joint
displacements .Thus,

Version 2 CE IIT, Kharagpur

 (R )
F
D 1 =
48 × 2 × 4 ⎛ 24 × 3 × 9 ⎞
+ ⎜− ⎟
16 ⎝ 36 ⎠
(1)

= 24 − 18 = 6 kN.m

(R )
F
D 2
= −24 kN.m

(R )
F
D 3 = 12 kN.m (2)

Thus,

⎧( RDF ) ⎫
⎪ 1⎪ ⎧6 ⎫
⎪ F ⎪ ⎪ ⎪
⎨( RD )2 ⎬ = ⎨−24 ⎬ (3)
⎪ F ⎪ ⎪12 ⎪
⎪⎩( RD )3 ⎪⎭ ⎩ ⎭
Next calculate stiffness coefficients. Apply unit rotation along u1 and calculate
reactions corresponding to the unknown joint displacements in the kinematically
determinate structure (vide Fig.23.5d)

Version 2 CE IIT, Kharagpur

Version 2 CE IIT, Kharagpur
Version 2 CE IIT, Kharagpur
 4 EI 4 EI
k11 = + = 1.667
4 6

2 EI
k 21 = = 0.5 EI
4

6 EI
k31 = − = −0.166 EI (4)
6× 6

Similarly, apply a unit rotation along u 2 and calculate reactions corresponding to

three degrees of freedom (see Fig.23.5e)

k12 = 0.5EI

k22 = EI

k32 = 0 (5)

Apply a unit displacement along u 3 and calculate joint reactions corresponding to

unknown displacements in the kinematically determinate structure.

Version 2 CE IIT, Kharagpur

 6 EI
k13 = − = −0.166 E
L2

k 23 = 0

12 EI
k33 = = 0.056 EI (6)
63

Finally applying the principle of superposition of joint forces, yields

⎧ F1 ⎫ ⎧6 ⎫ ⎧ 1.667 0.5 −0.166 ⎫ ⎧u1 ⎫

⎪ ⎪ ⎪ ⎪ ⎪ ⎪⎪ ⎪
⎨ F2 ⎬ = ⎨−24 ⎬ + EI ⎨ 0.5 1 0 ⎬ ⎨u2 ⎬
⎪ ⎪ ⎪ ⎪ ⎪−0.166 0 0.056 ⎪⎭ ⎪⎩u3 ⎪⎭
⎩ F3 ⎭ ⎩12 ⎭ ⎩
Now,

⎧ F1 ⎫ ⎧0⎫
⎪ ⎪ ⎪ ⎪
⎨ F2 ⎬ = ⎨0⎬ as there are no loads applied along u1 , u 2 and u 3 .Thus the
⎪ F ⎪ ⎪0⎪
⎩ 3⎭ ⎩ ⎭

unknown displacements are,

−1
⎧u1 ⎫ ⎡ 1 0.5 −0.166 ⎤ ⎧6 ⎫
⎪ ⎪ ⎢ ⎪ ⎪
0 ⎥⎥
1
⎨u2 ⎬ = − ⎢ 0.5 1 ⎨−24 ⎬ (7)
⎪u ⎪ EI ⎪ ⎪
⎩ 3⎭ ⎢⎣ −0.166 0 0.056 ⎥⎦ ⎩−24 ⎭
Solving

18.996
u1 =
EI

14.502
u2 =
EI

270.587
u3 = − (8)
EI

Version 2 CE IIT, Kharagpur

Summary
The flexibility coefficient and stiffness coefficients are defined in this section.
Construction of stiffness matrix for a simple member is explained. A few simple
problems are solved by the direct stiffness method. The difference between the
slope-deflection method and the direct stiffness method is clearly brought out.

Version 2 CE IIT, Kharagpur